Knee osteoarthritis (OA) affects 12% of older adults (4.3 million Americans) and contributes to diminished physical function, poor quality of life, and reduced life expectancy. However, the relationship between the structural joint damage that characterizes OA and its prominent clinical feature: pain, is poorly understood. In fact, of all people with radiographic evidence of OA, only about 50% have pain, and even people with severe OA often report being pain free. Meanwhile, many older adults with severe knee pain do not exhibit radiographic signs of OA. Our understanding of pain in knee OA has grown dramatically in recent years-we now know that structural damage to the knee joint is only one of many possible contributions to the experience of knee pain in older adults. Evidence suggests that psychological influences (such as depression, fear of pain, and pain catastrophizing) and central nervous system (CNS) influences (such as sensitization of the neural networks responsible for conjuring the pain experience) are important factors in painful knee OA. These variables have been studied in isolation, but we have yet to examine these variables concurrently in the same study population. We propose to conduct a two-part study to better characterize the pain experience in knee OA; part 1: establish the strength of the relationship between potential explanatory variables and pain severity in a population of older adults with knee OA, and part 2: determine whether these clinically measurable variables are representative of distinct pain phenotypes within knee OA. We propose to recruit 150 participants with symptomatic knee OA from orthopedic and primary care practices throughout the Denver area. Each participant will undergo a single testing session, completing a battery of clinically reproducible measures that are thought to influence pain in knee OA. The measures will be drawn from the following domains: 1) peripheral (knee) damage, 2) psychological distress and 3) CNS dysfunction. The variables will then be included in a multiple linear regression model to determine their relative contributions to pain severity. Based on the results of this analysis, measures will be chosen for a latent profile analysis to explore for homogenous subgroups within the heterogeneous knee OA population. We hypothesize that the proposed measures (or interactions between measures) will yield distinct phenotypes of knee OA, although the characteristics of such phenotypes are currently unknown. Thus, we anticipate this study will break new ground for future research, which may then seek to validate these pain phenotypes as clinically distinct entities, with different natural histories, prognoses, and potental routes of clinical care. Ultimately, our goal is to enable clinicians to better prioritize and targt interventions to individual pain phenotypes, thereby diminishing the cost and potentially iatrogenic complications of pharmaceutical and surgical interventions currently applied broadly across the diagnostic label of painful knee OA.